ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/CBOR-XS/XS.xs
Revision: 1.5
Committed: Sat Oct 26 21:14:20 2013 UTC (10 years, 6 months ago) by root
Branch: MAIN
Changes since 1.4: +11 -22 lines
Log Message:
*** empty log message ***

File Contents

# User Rev Content
1 root 1.1 #include "EXTERN.h"
2     #include "perl.h"
3     #include "XSUB.h"
4    
5     #include <assert.h>
6     #include <string.h>
7     #include <stdlib.h>
8     #include <stdio.h>
9     #include <limits.h>
10     #include <float.h>
11    
12     #include "ecb.h"
13    
14     #define F_SHRINK 0x00000200UL
15     #define F_ALLOW_UNKNOWN 0x00002000UL
16    
17     #define INIT_SIZE 32 // initial scalar size to be allocated
18    
19     #define SB do {
20     #define SE } while (0)
21    
22     #define IN_RANGE_INC(type,val,beg,end) \
23     ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
24     <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
25    
26     #define ERR_NESTING_EXCEEDED "cbor text or perl structure exceeds maximum nesting level (max_depth set too low?)"
27    
28     #ifdef USE_ITHREADS
29     # define CBOR_SLOW 1
30     # define CBOR_STASH (cbor_stash ? cbor_stash : gv_stashpv ("CBOR::XS", 1))
31     #else
32     # define CBOR_SLOW 0
33     # define CBOR_STASH cbor_stash
34     #endif
35    
36     static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS::
37     static SV *cbor_true, *cbor_false;
38    
39     typedef struct {
40     U32 flags;
41     U32 max_depth;
42     STRLEN max_size;
43    
44     SV *cb_object;
45     HV *cb_sk_object;
46     } CBOR;
47    
48 root 1.5 ecb_inline void
49 root 1.1 cbor_init (CBOR *cbor)
50     {
51     Zero (cbor, 1, CBOR);
52     cbor->max_depth = 512;
53     }
54    
55     /////////////////////////////////////////////////////////////////////////////
56     // utility functions
57    
58 root 1.5 ecb_inline SV *
59 root 1.1 get_bool (const char *name)
60     {
61     SV *sv = get_sv (name, 1);
62    
63     SvREADONLY_on (sv);
64     SvREADONLY_on (SvRV (sv));
65    
66     return sv;
67     }
68    
69 root 1.5 ecb_inline void
70 root 1.1 shrink (SV *sv)
71     {
72     sv_utf8_downgrade (sv, 1);
73    
74     if (SvLEN (sv) > SvCUR (sv) + 1)
75     {
76     #ifdef SvPV_shrink_to_cur
77     SvPV_shrink_to_cur (sv);
78     #elif defined (SvPV_renew)
79     SvPV_renew (sv, SvCUR (sv) + 1);
80     #endif
81     }
82     }
83    
84     /////////////////////////////////////////////////////////////////////////////
85     // fp hell
86    
87     //TODO
88    
89     /////////////////////////////////////////////////////////////////////////////
90     // encoder
91    
92     // structure used for encoding CBOR
93     typedef struct
94     {
95     char *cur; // SvPVX (sv) + current output position
96     char *end; // SvEND (sv)
97     SV *sv; // result scalar
98     CBOR cbor;
99     U32 depth; // recursion level
100     } enc_t;
101    
102 root 1.5 ecb_inline void
103 root 1.1 need (enc_t *enc, STRLEN len)
104     {
105 root 1.5 if (ecb_expect_false (enc->cur + len >= enc->end))
106 root 1.1 {
107     STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
108     SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
109     enc->cur = SvPVX (enc->sv) + cur;
110     enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
111     }
112     }
113    
114 root 1.5 ecb_inline void
115 root 1.1 encode_ch (enc_t *enc, char ch)
116     {
117     need (enc, 1);
118     *enc->cur++ = ch;
119     }
120    
121     static void
122     encode_uint (enc_t *enc, int major, UV len)
123     {
124     need (enc, 9);
125    
126     if (len < 24)
127     *enc->cur++ = major | len;
128 root 1.4 else if (len <= 0xff)
129 root 1.1 {
130     *enc->cur++ = major | 24;
131     *enc->cur++ = len;
132     }
133 root 1.4 else if (len <= 0xffff)
134 root 1.1 {
135     *enc->cur++ = major | 25;
136     *enc->cur++ = len >> 8;
137     *enc->cur++ = len;
138     }
139 root 1.4 else if (len <= 0xffffffff)
140 root 1.1 {
141     *enc->cur++ = major | 26;
142     *enc->cur++ = len >> 24;
143     *enc->cur++ = len >> 16;
144     *enc->cur++ = len >> 8;
145     *enc->cur++ = len;
146     }
147 root 1.4 else
148 root 1.1 {
149     *enc->cur++ = major | 27;
150     *enc->cur++ = len >> 56;
151     *enc->cur++ = len >> 48;
152     *enc->cur++ = len >> 40;
153     *enc->cur++ = len >> 32;
154     *enc->cur++ = len >> 24;
155     *enc->cur++ = len >> 16;
156     *enc->cur++ = len >> 8;
157     *enc->cur++ = len;
158     }
159     }
160    
161     static void
162     encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
163     {
164     encode_uint (enc, utf8 ? 0x60 : 0x40, len);
165     need (enc, len);
166     memcpy (enc->cur, str, len);
167     enc->cur += len;
168     }
169    
170     static void encode_sv (enc_t *enc, SV *sv);
171    
172     static void
173     encode_av (enc_t *enc, AV *av)
174     {
175     int i, len = av_len (av);
176    
177     if (enc->depth >= enc->cbor.max_depth)
178     croak (ERR_NESTING_EXCEEDED);
179    
180     ++enc->depth;
181    
182     encode_uint (enc, 0x80, len + 1);
183    
184     for (i = 0; i <= len; ++i)
185     {
186     SV **svp = av_fetch (av, i, 0);
187     encode_sv (enc, svp ? *svp : &PL_sv_undef);
188     }
189    
190     --enc->depth;
191     }
192    
193     static void
194     encode_hv (enc_t *enc, HV *hv)
195     {
196     HE *he;
197    
198     if (enc->depth >= enc->cbor.max_depth)
199     croak (ERR_NESTING_EXCEEDED);
200    
201     ++enc->depth;
202    
203     int pairs = hv_iterinit (hv);
204     int mg = SvMAGICAL (hv);
205    
206     if (mg)
207     encode_ch (enc, 0xa0 | 31);
208     else
209     encode_uint (enc, 0xa0, pairs);
210    
211     while ((he = hv_iternext (hv)))
212     {
213     if (HeKLEN (he) == HEf_SVKEY)
214     encode_sv (enc, HeSVKEY (he));
215     else
216     encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
217    
218 root 1.5 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
219 root 1.1 }
220    
221     if (mg)
222     encode_ch (enc, 0xe0 | 31);
223    
224     --enc->depth;
225     }
226    
227     // encode objects, arrays and special \0=false and \1=true values.
228     static void
229     encode_rv (enc_t *enc, SV *sv)
230     {
231     svtype svt;
232    
233     SvGETMAGIC (sv);
234     svt = SvTYPE (sv);
235    
236 root 1.5 if (ecb_expect_false (SvOBJECT (sv)))
237 root 1.1 {
238     HV *stash = !CBOR_SLOW || cbor_boolean_stash
239     ? cbor_boolean_stash
240     : gv_stashpv ("CBOR::XS::Boolean", 1);
241    
242     if (SvSTASH (sv) == stash)
243     encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);
244     else
245     {
246     #if 0 //TODO
247     if (enc->cbor.flags & F_CONV_BLESSED)
248     {
249     // we re-bless the reference to get overload and other niceties right
250     GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
251    
252     if (to_cbor)
253     {
254     dSP;
255    
256     ENTER; SAVETMPS; PUSHMARK (SP);
257     XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
258    
259     // calling with G_SCALAR ensures that we always get a 1 return value
260     PUTBACK;
261     call_sv ((SV *)GvCV (to_cbor), G_SCALAR);
262     SPAGAIN;
263    
264     // catch this surprisingly common error
265     if (SvROK (TOPs) && SvRV (TOPs) == sv)
266     croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
267    
268     sv = POPs;
269     PUTBACK;
270    
271     encode_sv (enc, sv);
272    
273     FREETMPS; LEAVE;
274     }
275     else if (enc->cbor.flags & F_ALLOW_BLESSED)
276     encode_str (enc, "null", 4, 0);
277     else
278     croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it",
279     SvPV_nolen (sv_2mortal (newRV_inc (sv))));
280     }
281     else if (enc->cbor.flags & F_ALLOW_BLESSED)
282     encode_str (enc, "null", 4, 0);
283     else
284     croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
285     SvPV_nolen (sv_2mortal (newRV_inc (sv))));
286     #endif
287     }
288     }
289     else if (svt == SVt_PVHV)
290     encode_hv (enc, (HV *)sv);
291     else if (svt == SVt_PVAV)
292     encode_av (enc, (AV *)sv);
293     else if (svt < SVt_PVAV)
294     {
295     STRLEN len = 0;
296     char *pv = svt ? SvPV (sv, len) : 0;
297    
298     if (len == 1 && *pv == '1')
299     encode_ch (enc, 0xe0 | 21);
300     else if (len == 1 && *pv == '0')
301     encode_ch (enc, 0xe0 | 20);
302     else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
303     encode_ch (enc, 0xe0 | 23);
304     else
305     croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
306     SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307     }
308     else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
309     encode_ch (enc, 0xe0 | 23);
310     else
311     croak ("encountered %s, but CBOR can only represent references to arrays or hashes",
312     SvPV_nolen (sv_2mortal (newRV_inc (sv))));
313     }
314    
315     static void
316     encode_nv (enc_t *enc, SV *sv)
317     {
318     double nv = SvNVX (sv);
319    
320     need (enc, 9);
321    
322 root 1.5 if (ecb_expect_false (nv == (U32)nv))
323 root 1.1 encode_uint (enc, 0x00, (U32)nv);
324     //TODO: maybe I32?
325 root 1.5 else if (ecb_expect_false (nv == (float)nv))
326 root 1.1 {
327     uint32_t fp = ecb_float_to_binary32 (nv);
328    
329     *enc->cur++ = 0xe0 | 26;
330    
331     if (!ecb_big_endian ())
332     fp = ecb_bswap32 (fp);
333    
334     memcpy (enc->cur, &fp, 4);
335     enc->cur += 4;
336     }
337     else
338     {
339     uint64_t fp = ecb_double_to_binary64 (nv);
340    
341     *enc->cur++ = 0xe0 | 27;
342    
343     if (!ecb_big_endian ())
344     fp = ecb_bswap64 (fp);
345    
346     memcpy (enc->cur, &fp, 8);
347     enc->cur += 8;
348     }
349     }
350    
351     static void
352     encode_sv (enc_t *enc, SV *sv)
353     {
354     SvGETMAGIC (sv);
355    
356     if (SvPOKp (sv))
357     {
358     STRLEN len;
359     char *str = SvPV (sv, len);
360     encode_str (enc, SvUTF8 (sv), str, len);
361     }
362     else if (SvNOKp (sv))
363     encode_nv (enc, sv);
364     else if (SvIOKp (sv))
365     {
366     if (SvIsUV (sv))
367     encode_uint (enc, 0x00, SvUVX (sv));
368     else if (SvIVX (sv) >= 0)
369     encode_uint (enc, 0x00, SvIVX (sv));
370     else
371     encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
372     }
373     else if (SvROK (sv))
374     encode_rv (enc, SvRV (sv));
375     else if (!SvOK (sv))
376     encode_ch (enc, 0xe0 | 22);
377     else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
378     encode_ch (enc, 0xe0 | 23);
379     else
380     croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
381     SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
382     }
383    
384     static SV *
385     encode_cbor (SV *scalar, CBOR *cbor)
386     {
387     enc_t enc;
388    
389     enc.cbor = *cbor;
390     enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
391     enc.cur = SvPVX (enc.sv);
392     enc.end = SvEND (enc.sv);
393     enc.depth = 0;
394    
395     SvPOK_only (enc.sv);
396     encode_sv (&enc, scalar);
397    
398     SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
399     *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
400    
401     if (enc.cbor.flags & F_SHRINK)
402     shrink (enc.sv);
403    
404     return enc.sv;
405     }
406    
407     /////////////////////////////////////////////////////////////////////////////
408     // decoder
409    
410     // structure used for decoding CBOR
411     typedef struct
412     {
413     U8 *cur; // current parser pointer
414     U8 *end; // end of input string
415     const char *err; // parse error, if != 0
416     CBOR cbor;
417     U32 depth; // recursion depth
418     U32 maxdepth; // recursion depth limit
419     } dec_t;
420    
421     #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
422    
423 root 1.5 #define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
424 root 1.1
425     #define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)
426     #define DEC_DEC_DEPTH --dec->depth
427    
428     static UV
429     decode_uint (dec_t *dec)
430     {
431     switch (*dec->cur & 31)
432     {
433     case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
434     case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
435     case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
436     return *dec->cur++ & 31;
437    
438     case 24:
439     WANT (2);
440     dec->cur += 2;
441     return dec->cur[-1];
442    
443     case 25:
444     WANT (3);
445     dec->cur += 3;
446     return (((UV)dec->cur[-2]) << 8)
447     | ((UV)dec->cur[-1]);
448    
449     case 26:
450     WANT (5);
451     dec->cur += 5;
452     return (((UV)dec->cur[-4]) << 24)
453     | (((UV)dec->cur[-3]) << 16)
454     | (((UV)dec->cur[-2]) << 8)
455     | ((UV)dec->cur[-1]);
456    
457     case 27:
458     WANT (9);
459     dec->cur += 9;
460     return (((UV)dec->cur[-8]) << 56)
461     | (((UV)dec->cur[-7]) << 48)
462     | (((UV)dec->cur[-6]) << 40)
463     | (((UV)dec->cur[-5]) << 32)
464     | (((UV)dec->cur[-4]) << 24)
465     | (((UV)dec->cur[-3]) << 16)
466     | (((UV)dec->cur[-2]) << 8)
467     | ((UV)dec->cur[-1]);
468    
469     default:
470     ERR ("corrupted CBOR data (unsupported integer minor encoding)");
471     }
472    
473     fail:
474     return 0;
475     }
476    
477     static SV *decode_sv (dec_t *dec);
478    
479     static SV *
480     decode_av (dec_t *dec)
481     {
482     AV *av = newAV ();
483    
484     DEC_INC_DEPTH;
485    
486     if ((*dec->cur & 31) == 31)
487     {
488     ++dec->cur;
489    
490     for (;;)
491     {
492     WANT (1);
493    
494 root 1.2 if (*dec->cur == (0xe0 | 31))
495 root 1.1 {
496     ++dec->cur;
497     break;
498     }
499    
500     av_push (av, decode_sv (dec));
501     }
502     }
503     else
504     {
505     int i, len = decode_uint (dec);
506    
507     av_fill (av, len - 1);
508    
509     for (i = 0; i < len; ++i)
510     AvARRAY (av)[i] = decode_sv (dec);
511     }
512    
513     DEC_DEC_DEPTH;
514     return newRV_noinc ((SV *)av);
515    
516     fail:
517     SvREFCNT_dec (av);
518     DEC_DEC_DEPTH;
519     return &PL_sv_undef;
520     }
521    
522     static SV *
523     decode_hv (dec_t *dec)
524     {
525     HV *hv = newHV ();
526    
527     DEC_INC_DEPTH;
528    
529     if ((*dec->cur & 31) == 31)
530     {
531     ++dec->cur;
532    
533     for (;;)
534     {
535     WANT (1);
536    
537 root 1.2 if (*dec->cur == (0xe0 | 31))
538 root 1.1 {
539     ++dec->cur;
540     break;
541     }
542    
543     SV *k = decode_sv (dec);
544     SV *v = decode_sv (dec);
545    
546     hv_store_ent (hv, k, v, 0);
547     }
548     }
549     else
550     {
551     int len = decode_uint (dec);
552    
553     while (len--)
554     {
555     SV *k = decode_sv (dec);
556     SV *v = decode_sv (dec);
557    
558     hv_store_ent (hv, k, v, 0);
559     }
560     }
561    
562     DEC_DEC_DEPTH;
563     return newRV_noinc ((SV *)hv);
564    
565     #if 0
566     SV *sv;
567     HV *hv = newHV ();
568    
569     DEC_INC_DEPTH;
570     decode_ws (dec);
571    
572     for (;;)
573     {
574     // heuristic: assume that
575     // a) decode_str + hv_store_ent are abysmally slow.
576     // b) most hash keys are short, simple ascii text.
577     // => try to "fast-match" such strings to avoid
578     // the overhead of decode_str + hv_store_ent.
579     {
580     SV *value;
581     char *p = dec->cur;
582     char *e = p + 24; // only try up to 24 bytes
583    
584     for (;;)
585     {
586     // the >= 0x80 is false on most architectures
587     if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
588     {
589     // slow path, back up and use decode_str
590     SV *key = decode_str (dec);
591     if (!key)
592     goto fail;
593    
594     decode_ws (dec); EXPECT_CH (':');
595    
596     decode_ws (dec);
597     value = decode_sv (dec);
598     if (!value)
599     {
600     SvREFCNT_dec (key);
601     goto fail;
602     }
603    
604     hv_store_ent (hv, key, value, 0);
605     SvREFCNT_dec (key);
606    
607     break;
608     }
609     else if (*p == '"')
610     {
611     // fast path, got a simple key
612     char *key = dec->cur;
613     int len = p - key;
614     dec->cur = p + 1;
615    
616     decode_ws (dec); EXPECT_CH (':');
617    
618     decode_ws (dec);
619     value = decode_sv (dec);
620     if (!value)
621     goto fail;
622    
623     hv_store (hv, key, len, value, 0);
624    
625     break;
626     }
627    
628     ++p;
629     }
630     }
631    
632     decode_ws (dec);
633    
634     if (*dec->cur == '}')
635     {
636     ++dec->cur;
637     break;
638     }
639    
640     if (*dec->cur != ',')
641     ERR (", or } expected while parsing object/hash");
642    
643     ++dec->cur;
644    
645     decode_ws (dec);
646    
647     if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
648     {
649     ++dec->cur;
650     break;
651     }
652     }
653    
654     DEC_DEC_DEPTH;
655     sv = newRV_noinc ((SV *)hv);
656    
657     // check filter callbacks
658     if (dec->cbor.flags & F_HOOK)
659     {
660     if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
661     {
662     HE *cb, *he;
663    
664     hv_iterinit (hv);
665     he = hv_iternext (hv);
666     hv_iterinit (hv);
667    
668     // the next line creates a mortal sv each time its called.
669     // might want to optimise this for common cases.
670     cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
671    
672     if (cb)
673     {
674     dSP;
675     int count;
676    
677     ENTER; SAVETMPS; PUSHMARK (SP);
678     XPUSHs (HeVAL (he));
679     sv_2mortal (sv);
680    
681     PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
682    
683     if (count == 1)
684     {
685     sv = newSVsv (POPs);
686     FREETMPS; LEAVE;
687     return sv;
688     }
689    
690     SvREFCNT_inc (sv);
691     FREETMPS; LEAVE;
692     }
693     }
694    
695     if (dec->cbor.cb_object)
696     {
697     dSP;
698     int count;
699    
700     ENTER; SAVETMPS; PUSHMARK (SP);
701     XPUSHs (sv_2mortal (sv));
702    
703     PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
704    
705     if (count == 1)
706     {
707     sv = newSVsv (POPs);
708     FREETMPS; LEAVE;
709     return sv;
710     }
711    
712     SvREFCNT_inc (sv);
713     FREETMPS; LEAVE;
714     }
715     }
716    
717     return sv;
718     #endif
719    
720     fail:
721     SvREFCNT_dec (hv);
722     DEC_DEC_DEPTH;
723     return &PL_sv_undef;
724     }
725    
726     static SV *
727     decode_str (dec_t *dec, int utf8)
728     {
729     SV *sv;
730    
731     if ((*dec->cur & 31) == 31)
732     {
733     ++dec->cur;
734    
735     sv = newSVpvn ("", 0);
736    
737     // not very fast, and certainly not robust against illegal input
738     for (;;)
739     {
740     WANT (1);
741    
742 root 1.2 if (*dec->cur == (0xe0 | 31))
743 root 1.1 {
744     ++dec->cur;
745     break;
746     }
747    
748     SV *sv2 = decode_sv (dec);
749     sv_catsv (sv, sv2);
750     }
751     }
752     else
753     {
754     STRLEN len = decode_uint (dec);
755    
756     WANT (len);
757     sv = newSVpvn (dec->cur, len);
758     dec->cur += len;
759     }
760    
761     if (utf8)
762     SvUTF8_on (sv);
763    
764     return sv;
765    
766     fail:
767     return &PL_sv_undef;
768     }
769    
770     static SV *
771 root 1.3 decode_tagged (dec_t *dec)
772     {
773     UV tag = decode_uint (dec);
774     SV *sv = decode_sv (dec);
775    
776     if (tag == 55799) // 2.4.5 Self-Describe CBOR
777     return sv;
778    
779     AV *av = newAV ();
780     av_push (av, newSVuv (tag));
781     av_push (av, sv);
782     return newRV_noinc ((SV *)av);
783     }
784    
785     static SV *
786 root 1.1 decode_sv (dec_t *dec)
787     {
788     WANT (1);
789    
790     switch (*dec->cur >> 5)
791     {
792     case 0: // unsigned int
793     //TODO: 64 bit values on 3 2bit perls
794     return newSVuv (decode_uint (dec));
795     case 1: // negative int
796     return newSViv (-1 - (IV)decode_uint (dec));
797     case 2: // octet string
798     return decode_str (dec, 0);
799     case 3: // utf-8 string
800     return decode_str (dec, 1);
801     case 4: // array
802     return decode_av (dec);
803     case 5: // map
804     return decode_hv (dec);
805     case 6: // tag
806 root 1.3 return decode_tagged (dec);
807 root 1.1 case 7: // misc
808     switch (*dec->cur++ & 31)
809     {
810     case 20:
811     #if CBOR_SLOW
812     cbor_false = get_bool ("CBOR::XS::false");
813     #endif
814     return newSVsv (cbor_false);
815     case 21:
816     #if CBOR_SLOW
817     cbor_true = get_bool ("CBOR::XS::true");
818     #endif
819     return newSVsv (cbor_true);
820     case 22:
821     return newSVsv (&PL_sv_undef);
822    
823     case 25:
824 root 1.2 {
825     WANT (2);
826    
827     uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
828     dec->cur += 2;
829    
830     return newSVnv (ecb_binary16_to_float (fp));
831     }
832 root 1.1
833     case 26:
834     {
835     uint32_t fp;
836     WANT (4);
837     memcpy (&fp, dec->cur, 4);
838     dec->cur += 4;
839    
840     if (!ecb_big_endian ())
841     fp = ecb_bswap32 (fp);
842    
843     return newSVnv (ecb_binary32_to_float (fp));
844     }
845    
846     case 27:
847     {
848     uint64_t fp;
849     WANT (8);
850     memcpy (&fp, dec->cur, 8);
851     dec->cur += 8;
852    
853     if (!ecb_big_endian ())
854     fp = ecb_bswap64 (fp);
855    
856     return newSVnv (ecb_binary64_to_double (fp));
857     }
858    
859     // 0..19 unassigned
860     // 24 reserved + unassigned (reserved values are not encodable)
861     default:
862     ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
863     }
864    
865     break;
866     }
867     #if 0
868     switch (*dec->cur)
869     {
870     //case '"': ++dec->cur; return decode_str (dec);
871     case '[': ++dec->cur; return decode_av (dec);
872     case '{': ++dec->cur; return decode_hv (dec);
873    
874     case '-':
875     case '0': case '1': case '2': case '3': case '4':
876     case '5': case '6': case '7': case '8': case '9':
877     //TODO return decode_num (dec);
878    
879     case 't':
880     if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
881     {
882     dec->cur += 4;
883     #if CBOR_SLOW
884     cbor_true = get_bool ("CBOR::XS::true");
885     #endif
886     return newSVsv (cbor_true);
887     }
888     else
889     ERR ("'true' expected");
890    
891     break;
892    
893     case 'f':
894     if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
895     {
896     dec->cur += 5;
897     #if CBOR_SLOW
898     cbor_false = get_bool ("CBOR::XS::false");
899     #endif
900     return newSVsv (cbor_false);
901     }
902     else
903     ERR ("'false' expected");
904    
905     break;
906    
907     case 'n':
908     if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
909     {
910     dec->cur += 4;
911     return newSVsv (&PL_sv_undef);
912     }
913     else
914     ERR ("'null' expected");
915    
916     break;
917    
918     default:
919     ERR ("malformed CBOR string, neither array, object, number, string or atom");
920     break;
921     }
922     #endif
923    
924     fail:
925     return &PL_sv_undef;
926     }
927    
928     static SV *
929     decode_cbor (SV *string, CBOR *cbor, char **offset_return)
930     {
931     dec_t dec;
932     SV *sv;
933    
934     /* work around bugs in 5.10 where manipulating magic values
935     * makes perl ignore the magic in subsequent accesses.
936     * also make a copy of non-PV values, to get them into a clean
937     * state (SvPV should do that, but it's buggy, see below).
938     */
939     /*SvGETMAGIC (string);*/
940     if (SvMAGICAL (string) || !SvPOK (string))
941     string = sv_2mortal (newSVsv (string));
942    
943     SvUPGRADE (string, SVt_PV);
944    
945     /* work around a bug in perl 5.10, which causes SvCUR to fail an
946     * assertion with -DDEBUGGING, although SvCUR is documented to
947     * return the xpv_cur field which certainly exists after upgrading.
948     * according to nicholas clark, calling SvPOK fixes this.
949     * But it doesn't fix it, so try another workaround, call SvPV_nolen
950     * and hope for the best.
951     * Damnit, SvPV_nolen still trips over yet another assertion. This
952     * assertion business is seriously broken, try yet another workaround
953     * for the broken -DDEBUGGING.
954     */
955     {
956     #ifdef DEBUGGING
957     STRLEN offset = SvOK (string) ? sv_len (string) : 0;
958     #else
959     STRLEN offset = SvCUR (string);
960     #endif
961    
962     if (offset > cbor->max_size && cbor->max_size)
963     croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
964     (unsigned long)SvCUR (string), (unsigned long)cbor->max_size);
965     }
966    
967     sv_utf8_downgrade (string, 0);
968    
969     dec.cbor = *cbor;
970     dec.cur = (U8 *)SvPVX (string);
971     dec.end = (U8 *)SvEND (string);
972     dec.err = 0;
973     dec.depth = 0;
974    
975     if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
976     ;//TODO dec.cbor.flags |= F_HOOK;
977    
978     sv = decode_sv (&dec);
979    
980     if (offset_return)
981     *offset_return = dec.cur;
982    
983     if (!(offset_return || !sv))
984 root 1.2 if (dec.cur != dec.end && !dec.err)
985     dec.err = "garbage after CBOR object";
986    
987     if (dec.err)
988 root 1.1 {
989 root 1.2 SvREFCNT_dec (sv);
990     croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur);
991 root 1.1 }
992    
993     sv = sv_2mortal (sv);
994    
995     return sv;
996     }
997    
998     /////////////////////////////////////////////////////////////////////////////
999     // XS interface functions
1000    
1001     MODULE = CBOR::XS PACKAGE = CBOR::XS
1002    
1003     BOOT:
1004     {
1005     cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1006     cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
1007    
1008     cbor_true = get_bool ("CBOR::XS::true");
1009     cbor_false = get_bool ("CBOR::XS::false");
1010     }
1011    
1012     PROTOTYPES: DISABLE
1013    
1014     void CLONE (...)
1015     CODE:
1016     cbor_stash = 0;
1017     cbor_boolean_stash = 0;
1018    
1019     void new (char *klass)
1020     PPCODE:
1021     {
1022     SV *pv = NEWSV (0, sizeof (CBOR));
1023     SvPOK_only (pv);
1024     cbor_init ((CBOR *)SvPVX (pv));
1025     XPUSHs (sv_2mortal (sv_bless (
1026     newRV_noinc (pv),
1027     strEQ (klass, "CBOR::XS") ? CBOR_STASH : gv_stashpv (klass, 1)
1028     )));
1029     }
1030    
1031     void shrink (CBOR *self, int enable = 1)
1032     ALIAS:
1033     shrink = F_SHRINK
1034     allow_unknown = F_ALLOW_UNKNOWN
1035     PPCODE:
1036     {
1037     if (enable)
1038     self->flags |= ix;
1039     else
1040     self->flags &= ~ix;
1041    
1042     XPUSHs (ST (0));
1043     }
1044    
1045     void get_shrink (CBOR *self)
1046     ALIAS:
1047     get_shrink = F_SHRINK
1048     get_allow_unknown = F_ALLOW_UNKNOWN
1049     PPCODE:
1050     XPUSHs (boolSV (self->flags & ix));
1051    
1052     void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1053     PPCODE:
1054     self->max_depth = max_depth;
1055     XPUSHs (ST (0));
1056    
1057     U32 get_max_depth (CBOR *self)
1058     CODE:
1059     RETVAL = self->max_depth;
1060     OUTPUT:
1061     RETVAL
1062    
1063     void max_size (CBOR *self, U32 max_size = 0)
1064     PPCODE:
1065     self->max_size = max_size;
1066     XPUSHs (ST (0));
1067    
1068     int get_max_size (CBOR *self)
1069     CODE:
1070     RETVAL = self->max_size;
1071     OUTPUT:
1072     RETVAL
1073    
1074     #if 0 //TODO
1075    
1076     void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1077     PPCODE:
1078     {
1079     SvREFCNT_dec (self->cb_object);
1080     self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1081    
1082     XPUSHs (ST (0));
1083     }
1084    
1085     void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef)
1086     PPCODE:
1087     {
1088     if (!self->cb_sk_object)
1089     self->cb_sk_object = newHV ();
1090    
1091     if (SvOK (cb))
1092     hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1093     else
1094     {
1095     hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1096    
1097     if (!HvKEYS (self->cb_sk_object))
1098     {
1099     SvREFCNT_dec (self->cb_sk_object);
1100     self->cb_sk_object = 0;
1101     }
1102     }
1103    
1104     XPUSHs (ST (0));
1105     }
1106    
1107     #endif
1108    
1109     void encode (CBOR *self, SV *scalar)
1110     PPCODE:
1111     PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1112     XPUSHs (scalar);
1113    
1114     void decode (CBOR *self, SV *cborstr)
1115     PPCODE:
1116     PUTBACK; cborstr = decode_cbor (cborstr, self, 0); SPAGAIN;
1117     XPUSHs (cborstr);
1118    
1119     void decode_prefix (CBOR *self, SV *cborstr)
1120     PPCODE:
1121     {
1122     SV *sv;
1123     char *offset;
1124     PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1125     EXTEND (SP, 2);
1126     PUSHs (sv);
1127     PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1128     }
1129    
1130     void DESTROY (CBOR *self)
1131     CODE:
1132     SvREFCNT_dec (self->cb_sk_object);
1133     SvREFCNT_dec (self->cb_object);
1134    
1135     PROTOTYPES: ENABLE
1136    
1137     void encode_cbor (SV *scalar)
1138     PPCODE:
1139     {
1140     CBOR cbor;
1141     cbor_init (&cbor);
1142     PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1143     XPUSHs (scalar);
1144     }
1145    
1146     void decode_cbor (SV *cborstr)
1147     PPCODE:
1148     {
1149     CBOR cbor;
1150     cbor_init (&cbor);
1151     PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1152     XPUSHs (cborstr);
1153     }
1154